Mitra

Mitra: Connecting Every Patient to the Right Care

Our Story _I grew up in a city in India which had access to some of the best healthcare facilities. Finding a doctor or hospital was often a matter of choice rather than availability.

That perspective changed when I volunteered in rural communities. I witnessed firsthand how difficult it was for patients and families to navigate the healthcare system. The challenge wasn't always the lack of care—it was knowing where the right care existed and how to reach it. Families often relied on word of mouth, incomplete information, or long journeys without knowing whether a facility could actually meet their needs.

I realized that access to healthcare is not just about hospitals and doctors; it's also about access to trustworthy information and confident decision-making.

Mitra was born from that realization. _

We wanted to create an AI-powered assistant that helps healthcare providers make referral decisions with transparency and evidence, ensuring that patients are connected to the right care destination faster. By combining governed healthcare data, explainable AI, and human-centered design, Mitra aims to reduce uncertainty, improve care coordination, and help bridge the gap between available care and accessible care.

Because every patient deserves not just healthcare—but a clear path to the right healthcare.

In plain language, a coordinator describes a patient's needs. Mitra evaluates thousands of facilities, scores the best options, surfaces supporting evidence, identifies risks and gaps, and explains exactly why each recommendation was made.

Built on Databricks, Mitra combines governed healthcare data, agentic AI, and explainable intelligence to help patients reach the right care faster, while empowering providers to make decisions with confidence.

Tap the mic, speak in any of 6 Indian languages — "emergency near Patna, which clinic is best reviewed?" The agent searches around ** 10,000 real facilities** by geography, pulls past referral feedback, checks live weather, and answers out loud — quoting the exact facility text behind every claim, with a string match score, and a "check before you trust" warning when evidence is weak. Every decision is saved back, so the system improves with each referral

How we built it? Mitra was built on Databricks as an agentic healthcare referral assistant that turns natural-language patient needs into evidence-backed facility recommendations. This data was organized and managed on Databricks to ensure scalability, security, and trust

On top of this foundation, we built an AI agent that understands a coordinator’s referral request in plain language, extracts key care requirements, searches across thousands of facilities, and scores the best matches. Mitra does not just return results—it explains them. Each recommendation includes why a facility matched, what evidence supports the decision, confidence signals, and what should still be verified. Also we have some enhanced features on the app like maps and showing distances along with multilanguage support given the diversity of India

The application was deployed as a Databricks App, providing an intuitive interface where healthcare coordinators can interact with the system, compare options, and make faster, more transparent referral decisions.

Challenges we ran into when we added some enhanced features on the app and deployed them to link to our app. But we loved working on it and we will keep to improvising it to provide the best care to patients

Accomplishments that we're proud of Built an AI-powered referral assistant that transforms complex healthcare referral requests into evidence-backed recommendations in seconds. The solution supports natural-language interactions in multiple languages, provides geospatial facility visualization through maps, and addresses a critical healthcare challenge where referral delays and uncertainty can directly impact patient outcomes.

What we learned Our biggest learning was that healthcare doesn't need AI that simply provides answers—it needs AI that delivers trusted, explainable recommendations and enables users to interact naturally in their own language.

What's next for Mitra Mitra is just the beginning. Our next goal is to evolve from a referral recommendation tool into an becoming a proactive healthcare coordination agent

In the near term, we plan to: Integrate real-time provider availability, capacity, and scheduling data to make recommendations even more actionable. Expand multilingual capabilities to improve accessibility for diverse patient populations. Incorporate patient preferences, insurance coverage, and transportation considerations into referral decisions. Enhance explainability with richer evidence trails and outcome-based insights.

Built With

• agents
• codex
• databricks
• lakebase